Forecasting the Days that Matter:  Identifying Windows of Opportunity to Improve High-impact Weather Forecasts in the National Blend of Models Era
Matthew Day, National Weather Service, Norman, OK
Todd Lindley, National Weather Service
Ryan Barnes, National Weather Service
 ,
 ,

Abstract
A blended consensus of numerical weather prediction (NWP) models, such as the National Blend of  Models (NBM), has been shown to be more skillful than any individual deterministic model. Considering only average performance, however, can mask potential windows of opportunity for meteorologists to add value to the forecast process when impactful weather is anticipated.  This study investigates NBM verification as a time series of daily mean absolute errors across the WFO Norman forecast area.  Periods of poor NBM verification relative to forecaster edits are highlighted, and commonalities in NBM failures are identified.  Preliminary results show that deviations from model blended predictions lead to improved weather forecasts for the days that truly matter most to public safety, including critical fire weather episodes, arctic air outbreaks, and winter precipitation events. Basic knowledge of weather regimes in which blended solutions fail can inform the development of tools needed to aid forecasters in identifying windows of opportunity for human intervention in the forecast process.